Topological Data Analysis of Urban Air Quality Monitoring Networks: Vietoris-Rips Simplicial Filtration, Persistent Homology, and Geostatistical Connectivity in a Bifurcated Riverine Airshed
Continuous ambient air quality networks in riverine megacities exhibit multi-scale spatial dependencies poorly captured by isotropic geostatistics. We establish a Topological Data Analysis (TDA) and spatial graph framework evaluating sensor network geometry across 13 regulatory stations bisecting the lower Hooghly River conurbation. Using spherical Haversine metrics and Gaussian radial basis functions, we benchmark the network against classical graph hierarchies (Delaunay, Gabriel, Relative Neighborhood, and Minimum Spanning Tree) alongside Ripley K and Besag L-functions. Vietoris-Rips simplicial filtration yields exact persistent homology invariants. Zero-dimensional homology (H0) reveals three-stage hierarchical percolation with conurbation consolidation at eps = 14.50 km (mean lifespan L0 = 4.16 km; persistent entropy E0 = 3.18 bits), verifying Kruskal/Boruvka duality with the MST edge spectrum (L = 49.88 km). One-dimensional homology (H1) isolates a transient trans-river 1-cycle (Delta eps = 25.1 m), confirmed via Monte Carlo coordinate perturbation (sigma_jitter in [10, 500] m) as an ephemeral coordinate discretization artifact. Grounding the filtration in empirical telemetry (PM2.5, PM10, NO2), Moran I(eps) transitions from localized clustering (I = +1.017 at eps = 2.0 km) to random expectation (I = -0.083) at the eps = 7.0 km consolidation scale, aligning with semivariogram practical range (a_prac = 5.29 km). Finally, on binary filtration graphs, we prove combinatorial Laplacian proportionality to Geary contiguity ratio C(eps), and formulate a topological embedding tensor H_TDA in R^(13x6) for topology-aware graph neural networks.
Authors
- Saptarshi Sadhu (ORCID: https://orcid.org/0009-0009-7505-1973)
Institutions
- Adamas University (IN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-20
- DOI
- https://doi.org/10.21203/rs.3.rs-11793674/v1
- Primary Topic
- Topological and Geometric Data Analysis
- Type
- preprint